Slackware current 64 bit

Six drives - four 1TB drives, and two older 160GB drives.

So I went to hibernate my system last night, and it locked up and never shut down. Eventually I had no choice to power cycle the box. When it came back up, it did not resume, indicating the attempt to hibernate had not been successful. Instead it just booted in the normal manner. Two of the drives could not be mounted, and fsck could not fix them. One could be mounted, and it had about 100GB that I could recover, the rest of the files had a zero length or did not exist.

Has anyone had any experiencing recovering from a botched attempt to hibernate?

Is hibernation not reliable under linux/Slack? I've learned the hard way not to hibernate windows boxes, but how about linux boxes?

I have no problems at all with hibernating my Slackware -current netbook.

You may want to tell us how exactly you hibernated this computer.
Hibernating works reliably if you set it up correctly.

Eric

I also haven't had any problems with my 64 -current laptop and hibernation either, and I hibernate if I intend on using my computer the next day. How long did you wait?

Then, I can just note that you are playing with that valuable data over an in-development version and you accepted to be beta-tester, right?

That do not look like some poor and innocent laptop, which is the usual target of hibernating, right?

Congratulation! You are just discovered that that super-fancy combination of (most likely) RAID0, LVM and XFS is not resistant to power failures and/or (unexpected) hard resets. So, you just lived with a ticking clock-bomb in your box, which just happened to boom it today...

I suggest you, next time when you play with these things, just use bare partitions with a good journaling FS, like EXT4FS or similar. And, if you want to test if your system will survive on a hibernation failure, just do an hard reset and see what happen.

Yup! The failed hibernation is just like an hard reset/power failure case, for the innocent filesystems currently mounted...

Sadly, I believe that there is nothing to do with Slackware, because The Hibernation is just a game between the Linux kernel and (VERY IMPORTANT!) your BIOS. And, we remember? that Slackware use just the vanilla kernel.

So, you have there the kernel developers and the hardware companies which do not offer (right) support for Linux , if you want to yell on someone.

Still, I believe that your system was wrong configured and that important loss of data, will happened at any power failure, anyway...

1 members found this post helpful.

Not had any problems here, but the lack of possible ways in which you can be helped posted here so far is dismally poor.

If I were you, I would run a live CD recovery software distribution and get back what I could. I'm not really able to recommend any particular one, as different people have different requirements, but some to try are:

Parted Magic
Partition Wizard
System Rescue CD
Trinity
Ultimate boot cd

Plus many others.

Good luck

1 members found this post helpful.

Firstly I might mention that I'm not yelling at anyone or looking to blame someone, I'm just curious as to others experience with hibernating linux. Secondly, I use -current because 1) it works 2) I like to play with bleeding edge 3) yes, I'm well aware of the potential of catastrophic failure 4) I keep my data backups safe and sound where any failure of an os or hardware won't keep me from restoring everything quickly and painlessly. 5) Being linux, it takes less than an hour to reload the OS and restore anything that needs to be restored as I keep my /home and /opt directories safe and sound on different drives, and everything nicely backed up somewhere else. - as opposed to the pain in the arse it is to reload Windows. Where is that key...why won't it activate..gotta call Microsoft *again* and try to convince them that I really did not pirate this copy of Windows that I paid good hard cash for...where is that office CD (don't care, I switched to LibreOffice a while back, good by MS Office forever AFAIC)

Loosing the contents of three 3GB drives is however a bit annoying. Fortunately, everything was safely backed up and it didn't take long to restore. I actually got lazy the night before and didn't back up my data, and lost about two paragraphs that I had typed in a document - that was all I had done that day.

And yes, this most definitely is not some boring laptop . My boring Toshiba laptop is the only windows based machine I have ever seen that would hibernate 100% and recover every time. This box has 6 drives - two or them old IDE 160GB drives that I play with or put an mysql server on for testing or whatever I happen to be playing with. 8Gb ram (wish I had 16gb), Phenom 6 core, whatever was latest and greatest six months ago. Wish I had 8 core - 6 is nice, but quickly becomes not enough.

So anyhow, I've been hibernating this box successfully for a month or so, and it has always worked flawlessly.

I set up a bigger swap file because it would not hibernate with the one I originally had, 512k. I added the appropriate restore line to lilo.conf. To hibernate I do this:

echo disk > /sys/power/state

And that nicely saves stuff to disk and powers off. It usually takes about 60 seconds to do this. Power up is clean and fast, and up until now has worked fine.

After a month of successful hibernation, I'm inclined to blame this failure on cosmic rays. There is a possibility that by using bleeding edge -current slackware 64 bit and kernel (3.3.6) that some odd bug popped up and bit me, but it's the price I'm willing to pay. I have found this box to be far more stable than my best Windows box ever was.

1 members found this post helpful.

that many drives sounds like a server. I'd never hibernate/sleep that kind of system.

10 plus years using mdadm, multiple systems, raid 5 or raid 1, plus ext3, even with a couple of lost drives, it hasn't let me down yet. very tolerant of hard shutdown or freezes.

i never use sleep on servers, or file storage systems. but i'm pretty sure if i did, and it went awry, that my setups would tolerate the abuse for a while.

d

I have taken away the sleep capability of my kernel because I don't use it.

I use an LVM encrypted system which hibernates and wakes up without problems. I just had to tell the initrd.gz file that I was using hibernation and pass an special parameter to the kernel. There is a nice guide somewhere in the Net. My only problem is that "resurrecting" is somehow sluggish, but at least it works.

It is a good idea to test Sleep/Hibernation on a computer without putting data at risk before actually start to use the unit for serious work.
Well, I have had unclean "umounts" when properly rebooting because the shutdown scripts did not handle some special situations. I had to rewrite part of the scripts to dismount the filesystems clearly. Maybe you are using a weird software configuration that is not well handled by your hibernate scripts?

What FS were you using for your drives might I ask?

One was ext3, the other two were ext2. I was eventually able to fully recover the ext3 partition - the ext2 partitions were toast. They have been ext2 forever, never got around to updating to ext3/4, but they don't really have anything important that can't be quickly restored so it was way low on my priority list.

I am running Slackware 13.37 and I have had problems with hibernate and suspend.

My laptop froze while going into suspend one day and I was forced to do a hard-reset. The system did not respond to ctrl+alt+del or the SysReq stuff. I never used suspend again.

Earlier today I had a kernel panic right after resuming from hibernate. SysReq worked this time, or so I think. The final step in the SysReq thing worked but the system had to recover journal when I booted up again and I don't know if that is normal. (I am doing alt+SysReq+REISUB, btw.) I don't plan on using hibernate again.

The computer is a ThinkPad X201i running 13.37 with an encrypted LVM. If anyone wants more details please ask me.

How many times have you successfully hibernated this machine?

I was using -current, but I'm back to using 13.37 because of a nasty bug introduced into -current a week ago that keep X from starting if you use ATI drivers. I've duplicated it on every machine I've tested it with. I noticed they just recently updated some xf86-video-* packages, but I haven't had time to test it to see if its fixed.

But I digress...I hibernated without problem on -current for quite a while until one day it hard locked on hibernate. I lost my ext2 partitions, and was able to fully recover my ext3 partition. It could have been cosmic rays, it could have been a bug in the kernel, it could have been who knows...I use 3.3.6 slightly customized, and it has been rock solid.

I'm thinking that if I really want to hibernate, I could umount my partitions, and upgrade my ext2 partitions to ext3/ext4. But if I'm going to umount my partitions to prevent data loss, I might as will do a normal shutdown, it kinda defeats the purpose.

I need to play with hibernate some more before I'm ready to rely on it. Once bitten, twice shy, ya know....

This is not a bug at all. The xserver in current was updated to 1.12 and the proprietary drivers from AMD just don't support this new version currently. Maybe in Catalist 12.5 or 6.

Probably around 40 times. So it's pretty good, overall. But it only takes one time to mess things up.

I am using the thinkpad-acpi stuff. I wonder if that had anything to do with it? Maybe I need to tell pm-hibernate to unload and re-load those modules?